Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
October 23, 2019Physics in Medicine and Biology

Computer-aided diagnosis system for breast ultrasound images using deep learning

View Full Paper
Ask AI
Bookmark
Share

Authors

HTHiroki TanakaTohoku UniversitySCShih-Wei ChiuTohoku University HospitalTWTakanori WatanabeNational Defense Medical College

Discussion

Loading...

Member takes

Overview

Diagnostic study demonstrates high classification accuracy in patients with breast masses, suggesting deep learning ensemble systems can enhance clinical cancer diagnosis.

Key Points

  • To develop a deep learning-based computer-aided diagnosis system to distinguish malignant from benign breast masses using multi-view ultrasound images and identify diagnostic visual regions.
  • Obtained ultrasound images of 1,536 pathologically confirmed breast masses (897 malignant and 639 benign) captured from multiple probe angles from a multicenter Japanese clinical trial.
  • Constructed an ensemble network combining fine-tuned VGG19 and ResNet152 architectures with data augmentation and multi-view mass-level classification.
  • Evaluated performance on an independent test set of 154 masses (77 malignant and 77 benign) and generated classification heat maps to examine regions of model attention.
  • The ensemble network achieved a sensitivity of 90.9% (95% CI, 84.5–97.3) and a specificity of 87.0% (95% CI, 79.5–94.5) on the independent test set.
  • The model attained an area under the curve (AUC) of 0.951 (95% CI, 0.916–0.987), outperforming the individual CNN architectures.
  • Heat map visualizations revealed that the neural networks relied primarily on regions surrounding the breast masses rather than the masses themselves for accurate classification.

Cite This Study

Tanaka et al. (2019) studied this question.

synapsesocial.com/papers/69fc2d648f981c7dacdd9424https://doi.org/10.1088/1361-6560/ab5093
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1A Hybrid CNN-Transformer Deep Learning Model for Differentiating Benign and Malignant Breast Tumors Using Multi-View Ultrasound Images.2026
  2. 2Breast cancer detection using deep convolutional neural networks and support vector machines2019 · 510 citations
  3. 3A Hybrid CNN-Transformer Deep Learning Model for Differentiating Benign and Malignant Breast Tumors Using Multi-View Ultrasound Images2025 · 1 citations
  4. 4Automated Classification and Segmentation and Feature Extraction from Breast Imaging Data2024 · 2 citations
  5. 5Enhancing Automated Breast Cancer Detection: A CNN-Driven Method for Multi-Modal Imaging Techniques2025 · 4 citations